This revised application is submitted in response to RFA CA-94-026, "Prevention Clinical Trials Utilizing Intermediate Endpoints and their Modulation by Chemopreventive Agents". We propose translational trials of topical all-trans-retinoic acid (atRA) as a chemopreventive for cervical carcinoma utilizing unique intermediate endpoints of reduction in the viral genome copies of HPV per cell coupled with a loss of viral mRNA message. The pilot study is designed to identify the optimal dose required to modulate biochemical endpoints and the intermediate endpoints. The specific aims are: 1) To determine if topical atRA reduces the number of viral genome copies of HPV per cell and inhibits HPV E6/E7 gene expression; 2) To determine if retinoic acid receptors (RAR) and/or retinoid X receptors (RXR) isoform expression change with topical atRA therapy. Twenty subjects with early cervical dysplasia, will be randomly assigned to one of three dose levels of atRA per two subgroups (CIN I and II) for two months of therapy. Total number of study participants in the pilot will be 206 allowing for 30% drop out rate. Subjects will undergo colposcopy with biopsy at baseline and three months after four day therapy with atRA. The dose chosen for the definitive trial will be the dose that has the best overall performance in biomarker modulation (HPV genome copy and E6/E7 expression). A definitive trial will be carried out in a placebo-controlled randomized format. The specific aims: l) To determine if topical atRA given in a randomized, placebo controlled, double blind study will cause significant increase in the regression rate of CIN I and II; 2) To determine if the regression of CIN I and II from atRA or placebo is reflected in the modulation of intermediate biological markers (viral genome copy number of HPV and HPV E6/E7 expression). 3) To determine if the regression of CIN I & II in response to atRA or placebo is associated with expression of specific RAR and/or RXR isoforms. 57 subjects per two subgroups (CIN I or II) will be randomly assigned to a placebo or atRA arm involving an initial therapy for four days and follow-up therapy at three and six months for two days. A total of 326 subjects will be enrolled allowing for a drop out rate of 30%. The subjects will undergo coloscopy with biopsy at baseline and at 12 months or at any point of removal from the study. The primary endpoint will be regression of CIN I or II to squamous metaplasia or normal tissue. The existence of an association between the biomarkers and the primary endpoint will be determined by univariant contrast between the biomarkers and the regression rates. The relative predictive ability of the markers will be assessed using multiple logistic regression along with other potential predictive variables including the expression of RAR and/or RXR.